APPENDIX B Sampling Summary I. Household Selection National Survey and Regional Oversamples For sampling telephone households, Gallup uses a stratified, list-assisted random-digit design. In order to avoid listing bias, a random digit procedure designed to provide representation of both listed and unlisted (including not-yetlisted) numbers is used. Minority Supplement In order to efficiently complete additional interviews among blacks and Hispanics, a supplemental sample was purchased from Survey Sampling, Incorporated. The design of these disproportionate samples involved analysis of 1990 Census racial and ethnic density by tract. Random regeneration of the last two digits of the seed numbers was used to create the final sample of telephone numbers. II. Respondent Selection An enumeration of the birthdays of all adults (age 18 and older) in the household was made. The adult with the most recent birthday was then selected. III. Weighting Algorithms Weight was used both to correct disproportionality imposed at the design stage, and minimize a variety of possible types of survey error that would effect surveybased estimates, including both random error and systematic biases. IV. Call Design The sample was administered using Gallup's computerized Phone Management system. This system is an automated sample server that distributes phone numbers to each interviewer according to a customized sample design. A complete connect Appendix B 51 ------------------------------------------------------------------------- history was recorded for each number. Numbers which were not resolved during an attempt were distributed according to a scheduled call design. Each of the randomly selected numbers received a minimum of five calls in order to resolve the number. Availability of Technical Appendix For researchers and others interested in more specific methodological background, a technical appendix is available by contacting The Partnership for Organ Donation. The technical appendix includes the actual survey instrument and specific information on sample development, respondent selection, weighting algorithms, minority and regional oversamples, and call design. 52 Appendix BReturn to survey index